SEMI‐ROLLED LEAF 10 stabilizes catalase isozyme B to regulate leaf morphology and thermotolerance in rice (Oryza sativa L.)

Summary

Plant architecture and stress tolerance play important roles in rice breeding. Specific leaf morphologies and ideal plant architecture can effectively improve both abiotic stress resistance and rice grain yield. However, the mechanism by which plants simultaneously regulate leaf morphogenesis and stress resistance remains elusive. Here, we report that SRL10, which encodes a double‐stranded RNA‐binding protein, regulates leaf morphology and thermotolerance in rice through alteration of microRNA biogenesis. The srl10 mutant had a semi‐rolled leaf phenotype and elevated sensitivity to high temperature. SRL10 directly interacted with catalase isozyme B (CATB), and the two proteins mutually increased one other's stability to enhance hydrogen peroxide (H₂O₂) scavenging, thereby contributing to thermotolerance. The natural Hap3 (AGC) type of SRL10 allele was found to be present in the majority of aus rice accessions, and was identified as a thermotolerant allele under high temperature stress in both the field and the growth chamber. Moreover, the seed‐setting rate was 3.19 times higher and grain yield per plant was 1.68 times higher in near‐isogenic line (NIL) carrying Hap3 allele compared to plants carrying Hap1 allele under heat stress. Collectively, these results reveal a new locus of interest and define a novel SRL10–CATB based regulatory mechanism for developing cultivars with high temperature tolerance and stable yield. Furthermore, our findings provide a theoretical basis for simultaneous breeding for plant architecture and stress resistance.

Introduction

Rice yield is closely linked to both plant architecture and stress resistance. The two ‘Green Revolutions’ and super‐rice breeding programs are based on continuous improvement of these two characteristics (Guo et al., 2020). To capitalize on available resources and maximize seed setting, plants make constantly adjustments in response to changes in environmental factors (Xu et al., 2021). Certain architectures, can predispose plants for success under specific conditions. For example, upright plant architecture facilitates high‐density planting (Tian et al., 2019), whereas a semi‐dwarf architecture can enhance lodging resistance and crop yield (Guo et al., 2021; Liu et al., 2018; Wang and Li, 2005). Appropriate leaf rolling not only contributes to enhanced light penetration into the canopy (Chen et al., 2019b; Sun et al., 2020; Xu et al., 2018) but also helps to improve root system activity and strengthen lodging resistance (Sun et al., 2020; Zhang et al., 2009; Zou et al., 2011). Extreme climate events have huge negative impact on global food production (Ray et al., 2015); for crops that are vulnerable to temperature fluctuations, precise control over heat resistance will be necessary in the future to minimize plants damage (Mittler et al., 2012). It is therefore of great practical significance for rice production to cultivate new germplasm resources with beneficial leaf morphologies and enhanced heat tolerance.

Stages of leaf development include leaf initiation, polarity establishment and maintenance, leaf flattening and intercalary growth (Du et al., 2018). Leaf morphology is determined by polarity along three axes: adaxial–abaxial, medial–lateral and apical–basal (Hasson et al., 2010; Moon and Hake, 2011). Establishment of leaf polarity is the result of complex interaction between regulatory networks involving phytohormone signalling pathways, transcription factors and microRNAs (miRNAs) (Wang et al., 2020a; Xu et al., 2018). Significant progress has been made in recent years towards the mechanisms that regulate leaf morphology; many relevant genes have been identified in rice, such as SHALLOT‐LIKE1 (SLL1) (Zhang et al., 2009) and SLL2 (Zhang et al., 2015b), ABAXIALLY CURLED LEAF 1 (ACL1) (Li et al., 2010), SEMI‐ROLLED LEAF 1 (SRL1) (Li et al., 2017; Xiang et al., 2012;) and SRL2 (Liu et al., 2016), Class III HD‐Zip gene HOMEODOMAIN CONTAINING PROTEIN 4 (OSHB4) (Li et al., 2016b ), ARGONAUTE‐1 (AGO1) ISOFORMS B (AGO1b) (Li et al., 2019), and members of the RICE OUTERMOST CELL‐SPECIFIC GENE (Roc) family Roc8 and Roc5 (Fang et al., 2021; Sun et al., 2020). Many genes have pleiotropic effects in regulating stress resistance, plant architecture and nutrient utilization; examples include PHOTO‐SENSITIVE LEAF ROLLING 1 (PSL1) (Zhang et al., 2021a), SLL1 (Zhang et al., 2009), DENSE PANICLE 1 (DEP1)/qLL9 (Fu et al., 2019; Huang et al., 2009), IDEAL PLANT ARCHITECTUTRE1 (IPA1) (Lu et al., 2013; Springer, 2010; Wang et al., 2021b) and DWARF 1 (D1)/LW5 (Fujisawa et al., 1999; Zhu et al., 2020). PSL1 is known to modify cell wall structure and drought tolerance in rice (Zhang et al., 2021a), whereas D1 is involved in a complex network regulating plant height, leaf size and abiotic stress responses (Fujisawa et al., 1999; Jangam et al., 2016; Zhu et al., 2020).

Heat stress causes different degrees of damage to protein, membrane, RNA and cytoskeleton, and changes the efficiency of intracellular reactions, leading to metabolic imbalances (Vu et al., 2019). Many genes regulated thermotolerance through complex regulatory networks involving the rapid plasma membrane (PM) sensing mechanism, Ca²⁺ signal transduction, reactive oxygen species (ROS) metabolism, post‐transcriptional RNA modifications, ubiquitinated protein degradation and the unfolded protein response (UPR) in the endoplasmic reticulum (ER) and cytosol (Mittler et al., 2012). Genes involved in these processes include SLENDER GUY 1 (SLG1) (Xu et al., 2020), ADAPTATION TO ENVIRONMENTAL TEMPERATURE 1 (AET1) (Chen et al., 2019a), the NOP/Sun family members NSUN2 (Tang et al., 2020), THERMO‐TOLERANCE 1 (TT1) (Li et al., 2015), TT2 (Kan et al., 2022) and HIGH TEMPERATURE SENSITIVE 1 (HTS1) (Chen et al., 2021). Although many regulatory genes that influence leaf morphology or thermotolerance have been identified, there are few reports detailing combined analyses of leaf morphology and thermotolerance. Identifying genes that simultaneously regulate both leaf morphology and stress resistance should clarify the mechanisms by which these regulatory pathways are integrated and provide new avenues for breeding high‐yield, high‐quality rice varieties in the future.

In plants, proteins that contain double‐stranded RNA‐binding motifs (dsRBMs) play integral roles in small RNA biosynthesis (Raghuram et al., 2015). Moreover, dsRBMs are also known to be important in plant development and defence (Bartel and Bartel, 2003; Hiraguri et al., 2005; Waterhouse et al., 2001). Twelve dsRBM‐containing proteins have been identified in rice, including eight double‐stranded RNA‐binding (DRB) proteins and four Dicer‐like (DCL) proteins. Functional analysis of DCLs (such as OsDCL1, OsDCL3b and OsDCL4) has suggested that dsRBM‐containing proteins have critical roles in leaf morphology (Liu et al., 2005), small RNA biogenesis (Song et al., 2012) and basal resistance against rice blast disease (Zhang et al., 2015a). However, few members of DRBs family have been functionally characterized in rice. Moreover, there have been no reports of DRBs simultaneously regulating of plant architecture and stress resistance.

In the present study, we identified a rice mutant that displayed increased temperature sensitivity and a semi‐rolled leaf phenotype. The causal gene, named SRL10, encoded a dsRBM‐containing protein and regulated miRNA biogenesis. SRL10 not only affected leaf morphology, but also increased thermotolerance by interacting with catalase isozyme B (CATB) and enhancing the H₂O₂ scavenging ability of CATB. Natural variations in SRL10 confer differential degrees of thermotolerance to japonica, indica and aus cultivars; the Hap3 (AGC) allele of SRL10, which was prevalent in the majority of aus rice, was identified as a thermotolerant haplotype. We therefore propose that manipulation of SRL10 has potential applications in super‐high‐yield rice breeding by simultaneously improving plant architecture and stress resistance.

Results

Phenotypic characterization of the srl10 mutant

The srl10 mutant, identified in a screen of an ethyl methanesulfonate (EMS) mutant library in the ‘Wuyunjing 7’ (WYJ7) background, was sensitive to heat stress (Figure 1c,d) and had adaxially rolled leaves throughout the entire growth period (Figure 1a, 7a,b). The leaf rolling index (LRI) of srl10 at the tillering stage was approximately 36.22%, whereas WYJ7 leaves were nearly flat, with an average LRI of 0.82% (Figure 1g). Moreover, srl10 plants had twice as many bulliform cells (BCs) as WYJ7, and the BC area was 2.5 times larger (Figure 1b,e,f). These results suggested that the increased BC number and size may have been responsible for the observed adaxial leaf rolling in srl10. In response to heat stress, both WYJ7 and srl10 at seedling stage or tillering stage showed leaf damage in the form of twisted leaf tips, but less change in LRI (Figure 1c,d,i). srl10 plants had a significantly lower survival rate than WYJ7 plants under heat stress (Figure 1h). In addition, transcriptomic analysis revealed 118 differentially expressed genes (DEGs) related to heat stress responses between WYJ7 and srl10; these DEGs were primarily enriched in functions related to protein processing in ER (Figure S1d,e). Moreover, SRL10 was strongly induced by heat and abscisic acid (ABA) treatment (Figure S1f,g), suggesting that SRL10 is involved in regulation of thermotolerance.

Figure 1.

Phenotypic comparison of WYJ7 and srl10. (a) Representative images of WYJ7 and srl10 transverse leaf sections (upper left), whole leaves (right) and whole plants (lower left) at the tillering stage. Scale bars = 1.5 cm, 10 cm and 20 cm respectively. (b) Paraffin sections of WYJ7 and srl10 leaves. Red arrows represent bulliform cells (BCs). Scale bar = 100 μm. (c) Representative WYJ7 and srl10 plants at seedling stage grown at 28 °C (left) or 42 °C (right), Scale bar = 2.5 cm. (d) Representative WYJ7 and srl10 leaves at tillering stage grown at 28 °C (left) or 42 °C (right), Scale bar = 1 cm. (e,f) The areas (e) and numbers (f) of bulliform cells in WYJ7 and srl10. (g) Leaf rolling index (LRI) of WYJ7 and srl10 at tillering stage. (h) Survival rates of WYJ7 and srl10 plants. (i) Leaf rolling index (LRI) of WYJ7 and srl10 leaves grown at 28 °C or 42 °C. Data are shown as mean ± standard deviation. Asterisks indicate significant difference based on the Student's t test: ** for P < 0.01; ns, not significant.

Figure 7.

Mutations of SRL10 and CATB affect the polarity development of rice leaves. (a) Representative whole plants and leaf cross‐sections of WYJ7, srl10, catb and srl10/catb. Scale bars = 20 cm and 0.8 cm respectively. (b) Representative leaves of WYJ7, srl10, catb and srl10/catb. Scale bar = 3 cm. (c) Histological analysis of leaf cross‐sections in WYJ7, srl10, catb and srl10/catb. Left, visible light. Middle, blue and red fluorescence, with blue indicating vascular bundle sheath, mechanical tissue and phloem, and red indicating mesophyll cells. Right, green fluorescence showing epidermal cells and mechanical tissues. (d) Chromatogram showing sequences of the wild type (WYJ7) and transgenic CATB knockout line (catb). (e) Leaf rolling index (LRI) values for WYJ7, srl10, catb and srl10/catb plants. (f) Bulliform cell (BC) areas in WYJ7, srl10, catb and srl10/catb plants. (g) Mesophyll cell areas in WYJ7, srl10, catb and srl10/catb plants. Data are shown as mean ± standard deviation. Significant differences between groups are marked with different letters (Duncan's multiple range test, P < 0.05).

Cell wall structure, stomata morphology, leaf water content and leaf water loss rate are all closely related to the transpiration rate (Zhang et al., 2015b; Zhang et al., 2021a). We therefore tested these parameters in WYJ7 and srl10 plants. Water loss rates in isolated leaves were significantly lower in srl10 compared to WYJ7 plants (Figure 2c); the leaf water content of srl10 was 71.02%, which was 1.76% higher than that of WYJ7 (Figure 2d). Moreover, the BC walls were 2.3 times thicker in srl10 than in WYJ7 (Figure 2a,e). This was consistent with the differences observed in levels of cell wall components (namely cellulose, hemicellulose and pectin) between WYJ7 and srl10 leaves (Figure 2f–h). In addition, stomatal apertures were markedly smaller in srl10 than WYJ7 (Figure 2b,i,j), consistent with observed decreases in transpiration rate and stomatal conductance in srl10 plants (Figure 2l,m). These results suggested that abnormal development of leaf histocytes leads to changes in leaf physiological characteristics. Furthermore, analysis of yield‐trait performance in WYJ7 and srl10 showed that loss of SRL10 function resulted in significant decreases in grain filling rate (Figure 3d), 1000‐grain weight (Figure 3e), grain yield (Figure 3f), number of secondary branches (Figure 3l), grain numbers per panicle (Figure 3j) and photosynthetic rate (Figure 2k). Taken together, these results demonstrated that SRL10 had pleiotropic effects on rice growth and development.

Figure 2.

Investigation of physiological characters of WYJ7 and srl10. (a) Micrographs showing BC walls in WYJ7 and srl10 leaves. Regions outlined in red are shown in greater detail at right. Blue arrows indicate BC walls. Scale bar = 2 μm (left) and 0.5 μm (right). (b) Micrographs of stomatal morphology in WYJ7 and srl10 leaves. Scale bar = 50 μm (left) and 10 μm (right). (c) Water loss rate of detached leaves in WYJ7 and srl10, n = 5. (d) Leaf water content of WYJ7 and srl10, n = 5. (e) Cells wall thickness of BC in leaves of WYJ7 and srl10. (f–h) The level of cell wall component in leaves of WYJ7 and srl10: (f) Cellulose content; (g) Hemicellulose content; (h) Pectin content, n = 5. (i,j) Stomatal density per mm² (i) and stomatal aperture (j) in leaves of WYJ7 and srl10. (k–n) Photosynthetic characteristics of WYJ7 and srl10. (k) Photosynthetic rate, (l) conductance to H₂O, (m) transpiration rate and (n) intercellular CO₂ concentration of WYJ7 and srl10, n = 10. Data are given as mean ± standard deviation. Asterisks indicate significant difference based on the Student's t test: * for P < 0.05; ** for P < 0.01; ns, not significant.

Figure 3.

Yield‐trait performance of WYJ7 and srl10. (a) Plant morphology of WYJ7 and srl10 at the mature stage, bar = 20 cm. (b) Spike morphology of WYJ7 and srl10. Scale bar = 3 cm. (c) Representative grain size (upper) and brown rice size (lower) of WYJ7 and srl10. Forty grains are shown per sample. Scale bar = 1 cm. (d) Grain filling dynamic of WYJ7 and srl10. (e–l) Statistical analysis of yield‐trait between WYJ7 and srl10: 1000‐grain weight (e), grain yield per plant (f), panicle length (g), numbers of effective panicles (h), seed‐setting rate (i), grain numbers per panicle (j), primary branch numbers per panicle (k) and secondary branch numbers per panicle (l) of WYJ7 and srl10. Data are given as mean ± standard deviation. Asterisks indicate significant difference based on the Student's t test: ** for P < 0.01; ns, not significant.

Map‐based cloning of SRL10

The semi‐rolled leaf phenotype was used as the mapping trait to identify the underlying genetic basis. Genetic analysis indicated that the semi‐rolled leaf phenotype of srl10 was determined by a single recessive nuclear gene (Table S1). We mapped the locus to a 27.04‐kb interval between the simple sequence repeat (SSR) markers C10‐2 and C10‐3. This interval contained two putative open reading frames (ORFs), LOC_Os10g33970 and LOC_Os10g33980 (Figure 4a). DNA sequence analysis of these putative ORFs revealed the presence of a single nucleotide substitution (C → T) in the 29th base after ATG in the first exon of LOC_Os10g33970 in srl10 compared to the wild type (Figure 4a,e), which resulted in an Ala‐to‐Val substitution. We therefore inferred that LOC_Os10g33970 was the gene controlling the srl10 mutant phenotype.

Figure 4.

Map‐based cloning of SRL10. (a) Mapping of SRL10. (b–d) Complementation tests of SRL10. Representative images of WYJ7, srl10 and SRL10‐COM (T₂‐generation) plants (b) and leaves (c) under standard growth conditions. Scale bars = 20 cm and 7 cm respectively. (d) WYJ7, srl10 and SRL10‐COM plants under standard growth conditions (upper) and heat stress (lower). Scale bar = 5 cm. (e) Chromatograms showing the sequences of WYJ7, srl10 and SRL10‐COM in the relevant region of LOC_Os10g33970. (f) Volcano plot of differentially expressed miRNAs between WYJ7 and srl10. (g) Heat map of differentially expressed miRNAs between WYJ7 and srl10. (h) Leaf rolling index (LRI) values of WYJ7, srl10 and SRL10‐COM. Data are given as mean ± standard deviation. Significant differences between groups are marked with different letters (Duncan's multiple range test, P < 0.05). (i) Survival rate of WYJ7, srl10 and SRL10‐COM after heat stress. Data are given as mean ± standard deviation. Significant differences between groups are marked with different letters (Duncan's multiple range test, P < 0.05). (j) Relative expression level of miR166m in leaves of WYJ7 and srl10. Data are given as mean ± standard deviation. Asterisks indicate significant difference based on the Student's t test: ** for P < 0.01.

To confirm this hypothesis, we carried out a complementation assay, expressing LOC_Os10g33970 in the srl10 background (Figure 4b). The LRI values and survival rates of SRL10‐complementation (SRL10‐COM) plants did not differ from the wild type (Figure 4b–d,h,i). These results showed that the normal functioning of LOC_Os10g33970 in srl10 recovered wild‐type characteristics in the srl10 background. Furthermore, we overexpressed LOC_Os10g33970 in the WYJ7 background and obtained three independent transgene‐positive T₂ lines. All three independent lines expressed LOC_Os10g33970 at significantly higher levels, and displayed abaxially rolled flag leaves with decreased BC number and area compared to WYJ7 (Figure 5b–d; Figure S2). Moreover, the LRI of abaxially rolled leaves increased along with SRL10 expression levels (Figure S2c,d).

Figure 5.

Functional verification of SRL10. (a–d) Overexpression tests of SRL10. (a) Representative images of WYJ7 (WT) and SRL10‐overexpression (OE) whole plants (upper left), leaves (right) and leaf cross‐sections (lower left). Scale bars = 20 cm, 2.5 cm, and 0.5 cm respectively. (b) Left, histological analysis of WT and OE leaf cross‐sections. Regions outlined in red are shown in greater detail at right. Red arrows indicate bulliform cells (BCs). Scale bars = 200 μm and 100 μm respectively. (c,d) BC areas (c) and BC numbers (d) in WT and OE leaves. (e–h) Phenotype analysis of SRL10 knockout (KO) lines. (e) Representative images of ‘Nipponbare’ (NPB/WT) and LOC_Os10g33970 (SRL10‐KO) mutant whole plants (upper left), leaves (right) and leaf cross‐sections (lower left). Scale bars = 20 cm, 2 cm, and 0.5 cm respectively. (f) Left, histological analysis of WT and KO leaf cross‐sections. Regions outlined in red are shown in greater detail at right. Red arrows indicate BCs. Scale bars = 200 μm and 100 μm respectively. (g, h) BC areas (g) and BC numbers (h) in NPB (WT) and SRL10‐KO (KO) leaves. (i) Subcellular localization of green fluorescent protein (GFP)‐SRL10 fusion protein in rice protoplasts. Scale bar = 5 μm. 4′,6‐diamidino‐2‐phenylindole (DAPI) staining was used to confirm the nuclear localization. (j) SRL10 promoter activity in multiple rice tissues as determined by β‐glucuronidase (GUS) assays. (k) Relative SRL10 expression levels in WYJ7 as measured with quantitative reverse transcription PCR (qRT‐PCR). Data are given as mean ± standard deviation. Asterisks indicate significant difference based on the Student's t test: * for P < 0.05; ** for P < 0.01; ns, not significant.

We then generated three independent LOC_Os10g33970 knockout lines in the Nipponbare background with CRISPR/Cas9 (Figure 5e–h; Figure S3). The three lines contained 1‐bp, 2‐bp and 4‐bp deletions respectively (Figure S3), in the third exon of LOC_Os10g33970, which resulted in frameshift mutations. All three lines showed semi‐rolled leaves and increased BC numbers (Figure 5f–h), a combination similar to our observations in srl10. Thus, the phenotypes of both overexpression and loss‐of‐function plants indicated that LOC_Os10g33970 was a key regulator of leaf rolling and that mutation of this gene caused the temperature‐sensitive and semi‐rolled leaf phenotypes of srl10. We therefore designated LOC_Os10g33970 as SRL10.

SRL10 consisted of three exons and two introns, and was predicted to encode a DRB protein. Phylogenetic analysis demonstrated high similarity between rice SRL10 and homologues in Zea mays (71.05% sequence similarity), Glycine max (65.23%), Arabidopsis thaliana (63.92%), Oryza brachyantha (92.73%), Sorghum bicolor (72.16%) and Hordeum vulgare (72.50%) (Figure S4a,b). These findings demonstrated that SRL10 was conserved between monocotyledonous and dicotyledonous angiosperms, and is therefore likely to have a fundamental function in plants. The Arabidopsis homologue, AtDRB2, is required for miRNA biogenesis (Eamens et al., 2012). We therefore hypothesized that SRL10 may be involved in regulation of miRNA biosynthesis in rice. To confirm this hypothesis, we performed small RNA‐seq analysis and found significant differences in accumination of miR166m, miR399d and miR399j between WYJ7 and srl10, consistent with the results of quantitative reverse transcription PCR (qRT‐PCR) (Figure 4f,g,j). While there was no significant difference in the accumination of miR166m between WYJ7 and SRL10‐OE (Figure S6a). These results suggested that loss of SRL10 function affected miRNA biosynthesis.

Next, a β‐glucuronidase (GUS) reporter system was used to identify tissues expressing SRL10. Strong GUS signals were observed in the seedlings, roots, stems, leaves, leaf sheaths and panicles (Figure 5j), consistent with SRL10 expression levels in those tissues measured with qRT‐PCR (Figure 5k). A green fluorescent protein (GFP)–SRL10 fusion system was then used to determine the subcellular localization of SRL10, and the fluorescent signal was primarily observed in the cytoplasm (Figure 5i). These results indicated that SRL10 was ubiquitously expressed in different tissues and SRL10 protein was primarily localized to the cytoplasm.

SRL10 interacted with CATB

To uncover the genetic pathway involved in SRL10 regulation, we performed a yeast two‐hybrid (Y2H) screen of a rice cDNA library. Surprisingly, CATB, one of the three CAT isoforms that contributes to CAT activity in plants (Matsumura et al., 2002), was shown to interact with SRL10 (Figure S5). In a domain deletion experiment in yeast, we found that the first dsRBM of SRL10 interacted weakly with CATB, whereas the second dsRBM showed a strong interaction with CATB (Figure 6a). The interaction between CATB and SRL10 was further confirmed using several methods: split‐luciferase complementation (SLC) (Figure 6b,c), co‐immunoprecipitation (Co‐IP) (Figure 6d), glutathione S‐transferase (GST) pull‐down (Figure 6e) and bimolecular fluorescence complementation (BiFC) assays (Figure S7). The results of all four assays demonstrated that SRL10 did interact with CATB.

Figure 6.

SRL10 interacts with CATB. (a) Yeast two‐hybrid assays showing interactions between SRL10 and CATB. Upper, SRL10 regions tested for interactions with CATB. Lower, co‐transformed yeast clones that grew on SD–Leu–Trp medium (left), SD–Leu–Trp–Ade–His medium (middle), and SD–Leu–Trp–Ade–His medium with 0.04 mg·mL⁻¹ X‐α‐Galactosidase (right). pGBKT7‐53 + pGADT7‐T and pGBKT7‐SRL10 + pGADT7 were used as the positive and negative controls respectively. (b,c) Interactions between SRL10 and CATB as detected by split‐luciferase (SLC) assays in Nicotiana benthamiana leaves. Red circles represent injection sites. (d) Co‐immunoprecipitation assay. SRL10‐GFP‐FLAG and CATB‐RFP‐HA were co‐expressed in N. benthamiana leaves. Molecular mass markers are shown (kDa). (e) His‐CATB, glutathione S‐transferase (GST) and GST‐SRL10 were purified with pull‐down assays and detected with anti‐His antibodies (His‐CATB) and anti‐GST antibodies (GST and GST‐SRL10). Molecular mass markers are shown (kDa).

We then disrupted the CATB gene in WYJ7 via CRISPR/Cas9 to generate a mutant line, catb, containing a 1‐bp insertion in the second exon of CATB that resulted in early termination of translation (Figure 7a,b,d). catb plants exhibited abaxially rolled flag leaves (Figure 7b). A srl10/catb double mutant was constructed by crossing srl10 and catb; double mutants had adaxially rolled leaves, similar to those of the srl10 single mutant. However, LRI values were lower in the srl10/catb double mutant than in srl10 single mutant plants (Figure 7e). This indicated that knocking out of CATB partially masked the srl10 mutant phenotype, reducing its severity. Further microscopic observation revealed that the BC area was significantly higher in srl10 and srl10/catb mutants than in WYJ7, whereas the mesophyll cell area was significantly less in srl10 than in WYJ7 and catb plants (Figure 7c,f,g).

To clarify the genetic relationship between SRL10 and CATB, we generated a CATB overexpression line in the srl10 background (pACTIN1::CATB srl10) and an SRL10 overexpression line in the catb background (pACTIN1::SRL10 catb). Overexpression of CATB in srl10 did not reverse the semi‐rolled leaf phenotype of srl10, and the leaves of pACTIN1::SRL10 catb plants were still abaxially curled (Figure 8). These results suggested that both SRL10 and CATB played critical roles in leaf morphogenesis.

Figure 8.

Complementation analysis of SRL10 in catb mutant and CATB in srl10 mutant. (a) Representative WYJ7, srl10 and pACTIN1::CATB srl10 leaves (upper) and leaf cross‐sections (lower). Scale bars = 2 cm and 0.3 cm respectively. (b) Representative WYJ7, catb and pACTIN1::SRL10 catb leaves (upper) and leaf cross‐sections (lower). Scale bars = 2 cm and 0.5 cm respectively. (c) Relative expression levels of CATB in WYJ7, srl10 and pACTIN1::CATB srl10 complementation plants. (d) LRI values for WYJ7, srl10 and pACTIN1::CATB srl10 complementation plants. (e) Relative expression levels of SRL10 in WYJ7, catb and pACTIN1::SRL10 catb complementation plants. (f) LRI values for WYJ7, catb and pACTIN1::SRL10 catb complementation plants. Data are shown as mean ± standard deviation. Significant differences between groups are marked with different letters (Duncan's multiple range test, P < 0.05).

Overexpression of SRL10 enhanced thermotolerance in rice via stabilizing CATB protein

To verify the role of SRL10 in thermotolerance, we exposed WYJ7, srl10 and SRL10‐overexpression (OE) plants to heat stress. Thereafter, we observed the degree of leaf damage and changes of photosynthetic parameters. At 42 °C, SRL10‐OE plants had the smallest reduction in Fv/Fm values and a significantly higher survival rate than WYJ7 and srl10 (Figure 9d,e). These results suggested that SRL10‐OE plants experienced the least heat‐induced photosystem damaged. At these temperatures, SRL10‐OE plants also had the smallest amount of visible damage, whereas srl10 plants suffered the most damage (Figure 9a,b). These findings were consistent with observed differences in level of H₂O₂ and malondialdehyde (MDA), a marker of cellular ROS damage, in leaves from each line (Figure 9f,g). This suggested that although heat stress induced ROS accumulation in all three lines, ROS accumulation was significantly lower in SRL10‐OE than in WYJ7 or srl10. In addition, CAT activity was significantly higher in SRL10‐OE than in WYJ7 and srl10 leaves under heat stress (Figure 9h), consistent with the higher expression levels of CATB in SRL10‐OE (Figure 9i,j). These results suggested that overexpression of SRL10 enhanced CAT activity under heat stress.

Figure 9.

Overexpression of SRL10 enhances thermotolerance in rice. (a) Representative images of WYJ7, srl10 and SRL10‐overexpression (OE) plants grown at 28 °C or 42 °C for 48 h. Scale bar = 10 cm. (b) Nitroblue tetrazolium (NBT) and 3, 3′‐diaminobenzidine (DAB) staining in WYJ7, srl10 and SRL10‐OE plants grown at 28 °C or 42 °C for 48 h. (c) Spike morphology of WYJ7, srl10 and SRL10‐OE under 42 °C for 7 d at booting stage. (d) Survival rates of WYJ7, srl10 and SRL10‐OE plants grown at 42 °C for 7 days. (e) F_v/F_m measurements of WYJ7, srl10 and SRL10‐OE plants grown at 42 °C for 48 h. (f–h) Malondialdehyde (MDA) content (f), H₂O₂ levels (g) and catalase activity (h) in leaves of WYJ7, srl10 and SRL10‐OE plants grown at 42 °C for 48 h. (i,j) Relative expression level of SRL10 (i) and CATB (j) in leaves of WYJ7, srl10 and SRL10‐OE plants grown at 42 °C for 48 h. (k) Seed‐setting rates of WYJ7, srl10 and SRL10‐OE plants in the booting stage after growth at 42 °C for 7 days. (l,m) Degradation assay (l) and degradation curve (m) of glutathione S‐transferase (GST)‐tagged SRL10 in the absence or presence of His‐tagged CATB. Equal starting amounts of total protein were used for the degradation reactions, as indicated by Ponceau S staining. (n,o) Degradation assay (n) and degradation curve (o) of His‐CATB in the absence or presence of GST‐SRL10. Equal starting amounts of total protein were used for each degradation reaction as indicated with Ponceau S staining. Data are shown as mean ± standard deviation. Significant differences between groups are marked with different letters (Duncan's multiple range test, P < 0.05).

Considering that the components in a protein complex may mutually regulate the stability of each unit (Bello et al., 2019), we conducted cell‐free degradation assay with His‐CATB and GST‐SRL10. The results showed that mixture of GST‐SRL10 and His‐CATB dramatically elongated the ‘half‐life’ time of the proteins compared to either sample alone in this in vitro assay, suggesting that the interaction helped to maintain the stability of each and would promoting normal protein functioning (Figure 9l–o). Increased CATB stability due to SRL10 binding may improve the CATB‐mediated H₂O₂ scavenging ability, and increased SRL10 stability due to CATB binding may be beneficial to plant growth and development processes that are regulated by this DRB protein. Rice is extremely sensitive to heat stress, particularly at the reproductive stage; in this phase, such stress causes declines in seed‐setting rate and yield (Xu et al., 2020). We therefore heat‐stressed WYJ7, srl10 and SRL10‐OE plants at the booting stage. Although heat stress significantly decreased the spikelet fertility in all three lines: the seed‐setting rate of SRL10‐OE plants was still significantly higher than that of WYJ7 and srl10 plants (by 2.41 times and 6.02 times respectively) at 42 °C (Figure 9c,k). Taken together, these results implied that SRL10 positively regulated rice thermotolerance by enhancing CAT activity and stabilizing CATB.

Haplotype analysis of SRL10

The results discussed above demonstrated a crucial role of SRL10 in rice leaf morphology and thermotolerance, we therefore performed haplotype analysis on 3025 rice accessions to identify alleles. We identified three single nucleotide polymorphisms (SNPs) within the SRL10 coding sequence (CDS): SNP1^A/G (18105284), SNP2^A/G (18104024) and SNP3^C/T (18103611) (Figure 10a). Based on these three SNPs, the rice accessions could be grouped into three main haplotypes: Hap1 (AAC), Hap2 (GGT) and Hap3 (AGC). Japonica rice almost exclusively carried Hap1 (AAC), indica rice carried both Hap2 (GGT) and Hap3 (AGC) and aus rice primarily carried Hap3 (AGC) (Figure 10b).

Figure 10.

Natural variations in SRL10 were associated with thermotolerance. (a) Three major haplotypes (Hap1–3) in the SRL10 coding region. (b) Distribution of Hap1–3 in three rice subgroups: japonica, indica and aus. Hap1, Hap2 and Hap3 are indicated with red, yellow and blue respectively. The number of accessions with each haplotype is shown in brackets (japonica/indica/aus). (c) Temperature changes in Fuyang District over 1 month during the reproductive growth stage in 2020 and 2021. (d,e) Statistical analysis of the seed‐setting rate of accessions carrying Hap1–3 in 2020 (d) and 2021 (e). (f,g) Statistical analysis of grain yield per plant for accessions carrying Hap1–3 in 2020 (f) and 2021 (g). (h,i) Seed‐setting rates of accessions carrying Hap1–3 after growth at 28 °C (h) or 42 °C (i) for 3 days at the booting stage. (j,k) Relative expression of SRL10 (j) and CATB (k) in accessions carrying Hap1–3 after growth at 42 °C for 48 h in the seedling stage. Data are given as mean ± standard deviation. Asterisks indicate significant difference based on the Student's t test: * for P < 0.05; ** for P < 0.01; ns, not significant.

Natural variations of SRL10 were associated with field thermotolerance in rice

We next carried out a series of experiments to clarify whether the haplotypes that we identified conferred varying levels of thermotolerance in rice. Heading dates were recorded for plants with different haplotypes (Table S6). Meteorological data from the summer of 2020 and 2021 in the Fuyang district of Hangzhou (Figure 10c), where the rice plants were grown, showed that plants experienced higher temperature at the booting stage of these accessions in 2020 than in 2021. The temperature difference between 2020 and 2021 led to difference in yield‐trait performance between 2 years; the seed‐setting rate and grain yield per plants with Hap3 (AGC) were significantly higher than those with Hap1 (AAC) (Figure 10d–g). This suggested that Hap3 (AGC) contributed to subspecies‐specific heat resistance.

We next applied heat stress (42 °C) to rice accessions with each of the three haplotypes at the booting stage in growth chamber conditions. Although spikelet fertility was significantly decreased in plants with all three haplotypes at 42 °C, seed‐setting rate was significantly higher for plants carrying Hap3 (AGC) allele than for those carrying Hap1 (AAC) or Hap2 (GGT) allele (4.92 times and 1.21 times higher respectively) (Figure 10h,i). Moreover, the relative expression levels of SRL10 and CATB were significantly higher in plants carrying Hap3 (AGC) allele than those carrying Hap1 (AAC) allele at 42 °C (Figure 10j,k).

Subsequently, we selected the near‐isogenic lines (NILs) NIL‐SRL10 ^Hap1 and NIL‐SRL10 ^Hap3, which carried Hap1 (AAC) and Hap3 (AGC) genotypes respectively. Under field conditions, NIL‐SRL10 ^Hap1 plants had significantly lower seed‐setting rate and grain number per plant compared to NIL‐SRL10 ^Hap3 plants (Figure 11a–d). Furthermore, for plants grown in a growth chamber at 42 °C, the seed‐setting rate and grain yield per plant were significantly higher in NIL‐SRL10 ^Hap3 plants than in NIL‐SRL10 ^Hap1 plants (3.18 times and 1.68 times higher respectively) (Figure 11e–i). Under heat stress, SRL10 and CATB were also expressed at lower levels in NIL‐SRL10 ^Hap1 than NIL‐SRL10 ^Hap3 plants at the seedling stage (Figure 11j,k). More importantly, SRL10 was more stable at 42 °C in NIL‐SRL10 ^Hap3 than in NIL‐SRL10 ^Hap1 plants after 30 min of heat exposure (Figure 11l,m). These results suggested that Hap3 (AGC) type of the SRL10 allele, which was present in the majority of aus rice cultivars, was highly associated with subspecies‐specific heat resistance; this demonstrated the important role of SRL10 in rice thermotolerance in nature.

Figure 11.

Analysis of thermotolerance in NIL‐SRL10 ^Hap3 and NIL‐SRL10 ^Hap1 plants. (a) Comparison of panicles from the near‐isogenic lines (NILs) NIL‐SRL10 ^Hap3 and NIL‐SRL10 ^Hap1. Scale bar = 3 cm. (b–d) Panicle length (b), grain numbers per panicle (c) and seed‐setting rate (d) of NIL‐SRL10 ^Hap3 and NIL‐SRL10 ^Hap1. (e) Spike morphology of NIL‐SRL10 ^Hap3 and NIL‐SRL10 ^Hap1 plants grown at 28 °C or 42 °C for 3 days at the booting stage. (f,g) Seed‐setting rate of NIL‐SRL10 ^Hap3 and NIL‐SRL10 ^Hap1 plants grown at 28 °C (f) or 42 °C (g) for 3 days at the booting stage. (h‐i) Grain yield per plant of NIL‐SRL10 ^Hap3 and NIL‐SRL10 ^Hap1 plants grown at 28 °C (h) or 42 °C (i) for 3 days at the booting stage. (j‐k) Relative expression levels of SRL10 (j) and CATB (k) in NIL‐SRL10 ^Hap3 and NIL‐SRL10 ^Hap1 plants grown at 28 °C or 42 °C for 48 h at the seedling stage. (l) Thermal stability analysis of SRL10 in vivo. Protein abundance was visualized via immune detection with anti‐SRL10 antibodies. Equal protein loading was confirmed with anti‐actin antibodies. (m) Degradation curve of SRL10 in NIL‐SRL10 ^Hap3 and NIL‐SRL10 ^Hap1 plants grown at 42 °C. Data are given as mean ± standard deviation. Asterisks indicate significant difference based on the Student's t test: * for P < 0.05; ** for P < 0.01; ns, not significant.

Discussion

SRL10 determines leaf morphology by regulating the development of bulliform cells

In this study, we identified SRL10 as a key regulator of leaf rolling through knockout and overexpression analysis. Compared to wild type, the knockout line of SRL10 had higher numbers and size of BCs, leading to adaxial leaf rolling. In contrast, overexpression of SRL10 led to decreased BC number and area, which resulted in abaxial leaf rolling. Our data therefore strongly support a negative role of SRL10 in the formation and development of BCs on adaxially leaf surface. Besides, SRL10 might also regulate sclerenchyma cell development; the disappearance of sclerenchyma cells on the adaxial surface caused BCs on both sides of the small vascular bundle to connect in srl10 and SRL10‐KO plants. BCs are highly linked to rolled leaf phenotype in rice (Xu et al., 2018; Zhang et al., 2015b) and defective development of BC in particular is responsible for the rolled leaf phenotype of many mutants (Zhou et al., 2018). We therefore hypothesized that adaxially rolled leaves in srl10 were primarily a result of changes in the number and area of BCs. Many genes that contribute to BC development have been identified, such as LATERAL ORGAN BOUNDARIES DOMAIN (LBD) gene OsLBD3‐7 ((Li et al., 2016a), ROP‐INTERACTING RECEPTOR‐LIKE KINASE 1 (OsRRK1) (Ma et al., 2017) and LEAF ROLLING RECEPTOR‐LIKE CYTOPLASMIC KINASE 1 (LRRK1) (Zhou et al., 2018). Thus, the identification and cloning of SRL10 conducted in the present study expands the known molecular regulatory network of leaf morphological development in rice.

SRL10 regulates both leaf morphology and thermotolerance in rice

Transcriptomic analysis and qRT‐PCR showed that DEGs related to leaf development between WYJ7 and srl10 were primarily enriched in phytohormone signal transduction pathways, including OsCOW1 and RL14 (Figure S1a–c). OsCOW1 encodes a member of the YUCCA protein family and affects water homoeostasis in rice (Woo et al., 2007). RL14 modulates leaf rolling by affecting water transport in leaves (Fang et al., 2012). Moreover, miR165/miR166 have prominent roles in abaxial leaf fate by targeting the HD‐ZIP III genes PHB, REV and PHV (Fouracre & Poethig, 2016; Nogueira et al., 2007; Sakaguchi & Watanabe, 2012; Zhang et al., 2018). The HD‐ZIP genes OSHB1/LF1 and OSHB4 regulate leaf development in an auxin‐dependent manner in rice (Li et al.,  2016b; Zhang et al., 2021b). Therefore, the differentially expression of miR166m and its target genes (OSHB1, OSHB2, OSHB4, YABBY5) in WYJ7 and srl10 indicated that SRL10 affects leaf polarity by participating in complex regulatory network involving auxin, miRNA and transcription factors.

Numerous studies have demonstrated that CATB functions as a key H₂O₂ scavenging enzyme (Ye et al., 2011; Zhang et al., 2016), but its role in leaf morphology has rarely been reported. In the present study, loss of CATB function was found to result in abaxially rolled flag leaves and up‐regulation of miR399 acumination (Figure S6b). In contrast, the srl10/catb double mutant exhibited adaxially rolled leaves, suggesting that CATB did control leaf morphology. Arabidopsis CATALASE2 have been reported to influence leaf morphology by changes of auxin levels (Gao et al., 2014). Notably, auxin can induce the production of ROS, in turn, ROS downstream product reactive carbonyl species (RCS) regulates auxin signal transduction in a feed‐forward manner (Biswas et al., 2019). Therefore, the complex crosstalk between auxin and H₂O₂ could be a reasonable explanation for the regulatory function of CATB on leaf morphology (Biswas et al., 2019; Li et al., 2021). However, the regulatory relationship between H₂O₂ and auxin in catb plants requires further investigation. We speculated that the loss of CATB function might result in changes of auxin level in vivo, thereby leading to changes in histocytes and rolled leaf in rice. In this study, the regulatory role of SRL10 on rolled leaf might be related to the development of bulliform cells regulated by miRNA‐auxin. Moreover, the interaction between SRL10 and CATB may further affect the crosstalk between H₂O₂ and auxin. Collectively, we speculated that the crosstalk of SRL10‐miRNA‐auxin and H₂O₂‐auxin jointly regulates the balance of auxin metabolism in vivo, endowing SRL10 with synergistic effect on leaf morphology and stress resistance.

ABA regulates stress resistance through modulation of stomatal aperture and leaf transpiration capacity (Li et al., 2020). We here found that SRL10 participates in heat stress response and ABA response. Furthermore, H₂O₂ is a critical signalling molecule in plants, and the balance between H₂O₂ generation and scavenging rates contributes to normal plant development and stress resistance (Zhang et al., 2016). Although CAT is the primary enzyme responsible for scavenging H₂O₂, its affinity for H₂O₂ is extremely low (Foyer et al., 2009; Mhamdi et al., 2012; Zhang et al., 2016). The substrate channelling mechanism used in processes such as ROS scavenging suggested that binding of enzymes to other substances can greatly accelerate the reaction rate (Singleton et al., 2014; Zhang, 2011; Zhang et al., 2016). Thus, interaction between SRL10 and CATB may affect the affinity of CAT for H₂O₂, altering its H₂O₂ scavenging capacity. We observed significantly higher CAT activity and CATB expression levels in the leaves of SRL10‐OE plants, which led to lower H₂O₂ levels. This suggested that overexpression of SRL10 enhanced CAT activity under heat stress. Moreover, SRL10–CATB interactions contributed to the stability of both proteins. We therefore speculate that SRL10 participates in H₂O₂ metabolism, positively regulating thermotolerance by interacting with and stabilizing CATB.

The role of SRL10 in regulating thermotolerance may also be related to complex miRNA‐mediated regulatory networks. In Arabidopsis, AtDRB2 is required for appropriate regulation of the miRNA399/PHOSPHATE2 expression module (Pegler et al., 2019), which alters the salt stress response (Pegler et al., 2020). This suggests that differential expression of miR399 in WYJ7 and srl10 may also be related to thermotolerance. Furthermore, high temperature‐triggered increase in transpiration rate was considered as a strategy of enabling ‘heat avoidance’ by lowing leaf temperature (Aparecido et al., 2020; Haddad et al., 2021; Lin et al., 2017; Crawford et al., 2012). The complex relationships between cell wall structure, stomatal morphology, leaf water balance and the transpiration rate could underlie srl10 sensitivity to high temperature. The role of SRL10 in controlling thermotolerance appears to be complex and regulated by miRNA, ROS metabolism and other unknown factors. The details of these regulatory mechanisms are not yet clear, demonstrating a need for further in‐depth research in the future. Identifying new regulatory genes downstream of SRL10 should be focus of subsequent studies; this would further clarify the role of SRL10 in leaf morphology and thermotolerance. Our results serve as a reference for gene discovery and simultaneous molecular breeding of rice thermotolerance and leaf morphology.

SRL10 has pleiotropic effects on rice growth and development

Most dsRBM‐containing proteins have a second functional or catalytic domain, suggesting that such proteins participate in multiple biological processes such as miRNA biosynthesis (Burd and Dreyfuss, 1994), abiotic stress responses (Raghuram et al., 2015) and hormone signalling (Lu and Fedoroff, 2000). SRL10 contains two dsRBMs and a putative PHA03247 superfamily UL36 domain. We here found that a C to T substitution in the first exon of SRL10 led to an amino acid change in the first dsRBMs of SRL10 and was associated with the adaxial leaf rolling phenotype in srl10. A complementation assay showed that expression of LOC_Os10g33970 in srl10 rescued the semi‐rolled leaf and temperature‐sensitive phenotypes.

Interaction between SRL10 and CATB were verified both in vitro and in vivo with Y2H, Co‐IP, SLC, GST pull‐down and BiFC assays. We also confirmed that CATB‐GFP was localized to the peroxisome (Figure S7). In addition, we repeated the BiFC assay confirming the interaction between CATB and SRL10 with multiple organelle‐specific fluorescent markers (such as peroxisomes, ER, mitochondria and Golgi apparatus) and strong yellow fluorescent protein (YFP) signal was consistently observed, indicating interactions between SRL10 and CATB. However, there was no overlap observed between the YFP signal and any of the organellar markers. This might have been due to the complexity of protein expression (Tanz et al., 2013) or an unknown interaction mechanism. For example, some active proteins exist in multiple organelles and are shuttled between five or six subcellular structures (Carrie and Small, 2012; Small et al., 1998). Gao et al. (2021) found that co‐expression of ROD1 and CATB changed the localization of CATB. The details of interactions between CATB and SRL10 therefore require further investigation.

Most of the known proteins identified as SRL10 interactors here participate in multiple biological processes. For example, rice snf1‐related protein kinase 1 (SnRK1) family members SnRK1A mediates glucose metabolism (Lu et al., 2007); VASCULAR PLANT ONE‐ZINC FINGER 1 (VOZ1) mediates immune response (Wang et al., 2021a); and AUTHENTIC HIS PHOSPHOTRANSFER PROTEIN 2 (AHP2) mediates drought and salt resistance (Figure S5) (Sun et al., 2014). Moreover, the loss of SRL10 function resulted in open glumes, imperfect grain filling and reduced 1000‐grain weight. We therefore hypothesize that SRL10 has pleiotropic functions, including regulation of plant growth and resistance to environmental stresses.

Prospects for applying SRL10 in rice molecular breeding for stress resistance and high yield

Ideal plant architecture and strong stress tolerance are the two major targets of high‐yield breeding. Although many high temperature‐tolerant germplasm resources have been identified, few varieties have strong thermotolerance without associated yield decreases. We here confirmed that natural variation of SRL10 associated with field thermotolerance by analysing yield‐trait performance and SRL10 thermal stability in accessions with different SRL10 haplotypes (NIL‐SRL10 ^Hap1 and NIL‐SRL10 ^Hap3) under heat stress (42 °C). These results suggested that advantages in heat resistance were conferred by Hap3 (AGC) type of SRL10 allele, which resulted from higher SRL10 expression levels and protein stability; this minimized heat‐induced plant damage without decreasing yield. Because elite rice varieties containing Hap3 (AGC) exhibit stronger thermotolerance without an obvious decline in productivity, this natural variation of SRL10 constitutes an elite allelic variation for high‐yield breeding through ideal plant architecture and high thermotolerance. Collectively, our combined analysis of leaf morphology and thermotolerance revealed that integration of favourable SRL10 alleles could accelerate inter‐regional exchange of high‐performance rice germplasm resources, thus promoting sustainable development in global agriculture.

Materials and methods

Plant materials and growth conditions

Oryza sativa ssp. japonica variety WYJ7 seeds were mutagenized with a 1% EMS solution. Rice plants were grown in experimental fields at the China National Rice Research Institute in Fuyang District (Zhejiang Province, China) (30°4′52′′ E, 119°55′54′′ N) under natural environmental conditions. A forward genetic screen was then conducted for high temperature sensitivity and altered leaf morphology.

To test heat tolerance, seedlings were grown in growth chambers under control temperature conditions (28 °C/25 °C day/night temperature) or heat stress condition (42/25 °C day/night temperature) at the three‐leaf and five‐leaf stages. Plants were cultured in nutrient soil and hydroponic media, respectively, at 70%–80% relative humidity with a 14/10 h light/dark period. Seedlings cultured in hydroponic media were also treated with 50 μm ABA for 12 h at the five‐leaf stage.

For heat stress treatment, seedlings were cultured in paddy soil under a natural environment. At the booting stage, rice plants were divided into two groups and moved into separate plant growth chambers. One group of rice plants was grown under heat stress condition (42 °C from 10:00 to 16:00/28 °C from 16:01 to 9:59 for 3 or 7 days). The other group served as the control (28 °C from 10:00 to 16:00/25 °C from 16:01 to 9:59 for 3 or 7 days). Both groups were maintained with 70%–80% relative humidity and 800 μmol · m⁻² · s⁻¹ of light.

Map‐based cloning

To map the SRL10 locus, an F₂ population derived from a cross between srl10 and TN1 and newly developed simple sequence repeat (SSR) or sequence‐tagged site (STS) markers were used (Table S2). To identify the mutation site, genomic DNA fragments of candidate genes were amplified (Table S3) from WYJ7 and srl10, sequenced, and compared using SeqMAN (DNASTAR).

Constructs for transgenic plants

To construct the genomic DNA complementation vector, the 8796‐bp SRL10 genomic DNA sequence (including 2109‐bp upstream of the start codon, 5394‐bp coding region of SRL10 and 1293‐bp downstream of the stop codon) was amplified from WYJ7 and was cloned into the binary vector pCAMBIA1300. To construct the SRL10 and CATB CRISPR/Cas9 vectors, the target sequences of single guide RNAs (sgRNAs) were designed and then the OsU3‐SRL10 and OsU3‐CATB sgRNA expression cassettes were assemble in pYLCRISPR/Cas9‐MH (Ma et al., 2015). To construct the overexpression vector (pUbi::attR‐SRL10‐GFP‐3 × FLAG) of SRL10, the full‐length coding sequence of SRL10 was amplified from Nipponbare, then firstly cloned into the pDONR ZEO (Invitrogen), and recombined into the pUbi::attR‐GFP‐3 × FLAG. To generate pACTIN1::SRL10 catb and pACTIN1::CATB srl10 complementation lines, the full‐length coding region of SRL10 or CATB was inserted into the pCAMBIA2300‐ACTIN1 vector. All resulting constructs were transformed into rice calli via Agrobacterium tumefaciens‐mediated transformation (Toki et al., 2006). All the primers used in plasmid construction are shown in Table S4, and all of the constructs were confirmed by sequencing.

Phylogenetic analysis

Protein sequences of SRL10 were obtained from Rice Genome Annotation Project (http://rice.uga.edu/index.shtml). The sequences used in the phylogenetic analysis were obtained by a BLASTP search using the amino acid sequence of SRL10 as the query at the National Center for Biotechnology Information (NCBI, http://www.ncbi.nlm.nih.gov/). Multiple sequence alignments of protein were done using the DNAMAN program. A phylogenetic tree of aligned sequence was constructed in MEGA software with the bootstrap method and 1000 replicates (Gao et al., 2021; Qiu et al., 2018).

Histological promotor‐GUS assay

To construct the SRL10 promoter‐driven GUS reporter gene, a 2109‐bp fragment upstream of the SRL10 start codon (Table S4) was amplified and cloned into the EcoRI and NcoI sites of the binary vector pCAMBIA1305.1. The recombinant vector was then introduced into WYJ7 calli to generate transgenic plants. Several tissues from SRL10 _promoter :GUS transgenic rice was stained for GUS activity as previously described (Jefferson et al., 1987).

Subcellular localization of SRL10

For detection of the subcellular localization of SRL10, the full‐length cDNA of SRL10 amplified (primers used are listed in Table S4) from Nipponbare rice was cloned into the pBeacon‐NeGFP vector using the Gateway cloning system (Invitrogen). The result construct GFP‐SRL10 was transformed into rice protoplasts by the polyethylene glycol (PEG)‐mediated transformation method (Yu et al., 2014). Green fluorescent protein (GFP) signals were observed by laser scanning confocal microscopy (Zeiss LSM 700).

Measurement of physiological indices

The leaf rolling index (LRI) of rice plants was measured according to the method described by Xiang et al. (2012). The distance of leaf blade margins in the natural state (Ln) and in the unfolding state (Lw) in different lines was measured. LRI was then calculated with the following formula: LRI (%) = (Lw‐Ln) × 100/Lw. Fresh leaves of WYJ7 and srl10 were weighed at specified times to determine the water loss rate of detached blade. Leaf water content was determined as described previously (Qiu et al., 2019).

ROS‐related measurements

CAT activity, H₂O₂ and MDA content were quantified in leaves grown under heat stress for 48 h according to the manufacturer's instructions by using corresponding kits from Geruisi (http://www.geruisi‐bio.com/). Nitro blue tetrazolium (NBT) staining and 3, 3′‐diaminobenzidine (DAB) staining were performed as previously described (Thordal‐Christensen et al., 1997).

Histology and microscopic observations

For freehand section, the leaves were sliced into thin slice by a double blade, and then were observed and photographed using a fluorescence microscope (Leica DM4 B). Paraffin section analysis was performed by using the middle part of each leaf at seedling stage as previously described by Ruan et al. (2020). For frozen cross‐section assays, the leaves were immersed in the frozen embedding agent (Tissue‐Tek® O.C.T. Compound, SAKURA, JAPAN) for 2–3 h at −20 °C. Sections (15 μm) were cut with a freezing microtome (Leica CM1950). Slices were observed and photographed using a microscope (Leica DM4 B). Cellulose, hemicellulose, and pectin levels were measured in WYJ7 and srl10 according to Zhong and Läuchli (1993). The youngest fully expanded leaves of the rice plants were used for analysis of stomatal index and cell wall as described previously (Zhang et al., 2021). The areas of BCs, number of stomata, and stomatal aperture were calculated using Image J software (Ma et al., 2017).

Measurement of photosynthetic parameters

The photosynthetic parameters of WYJ7 and srl10 were measured at the tillering stage with a Li‐COR 6400 portable system according to the method described by Li et al. (2020).

Determination of agronomic traits

Agronomic traits such as effective panicle number, numbers of branches, grain numbers per panicle, seed‐setting rate, grain yield per plant and 1000‐grain weight were measured at the mature stage. To measure the grain filling rate, representative samples from the main panicle at 3, 6, 9, 12, 15, 19, 21, 24 and 27 d after flowering were used to measure the dry weight.

RNA extraction and quantitative real‐time PCR

Total RNA was extracted using the Total RNA Miniprep kit (Axygen, China) according to the manufacturer's instruction. First‐strand cDNA and the miRNA first‐strand cDNA were synthesized using the ReverTra Ace qPCR‐RT kit (Toyobo, Japan) and miRNA 1st Strand cDNA Synthesis Kit (by stem‐loop) (Vazyme, https://www.vazyme.com/Home.html) according to the user's manual respectively. Quantitative real‐time PCR (qRT‐PCR) analyses were performed using SYBR Premix Ex Taq (Takara, Japan) and gene‐specific primers (Table S5) on a CFX96TM Real‐Time System with snRNA U6 or Ubiquitin (UBQ) as an internal control.

Yeast two‐hybrid assays

In yeast two‐hybrid (Y2H) assays, for DNA‐binding domain (BD)‐fused SRL10, the full‐length coding sequences of SRL10 (Table S4) were fused in‐frame to the GAL4 BD in the pGBKT7 bait vector. For activation domain (AD)‐fused CATB, the full‐length coding sequences of CATB (Table S4) were fused in‐frame to the GAL4 AD in the pGADT7 prey vector. For a domain deletion experiment, we divided the full‐length coding sequences of SRL10 into three segments and they were cloned into pGBKT7 respectively. Different combinations of bait and prey constructs were co‐transformed into the yeast strain Y2H Gold as described previously (Ruan et al., 2020).

Bimolecular fluorescence complementation assay

The cDNA of SRL10 was cloned into pSAT1A‐cEYFP‐N1 (pE3080) for C‐terminal fusion and the cDNA of CATB was cloned into pSAT1A‐nEYFP‐N1 (pE3079) for N‐terminal fusion (primers used are listed in Table S4). YN + YC‐SRL10 and YN‐CATB + YC served as controls. YN‐CATB and YC‐SRL10 were then co‐transformed pair‐wise into protoplasts by PEG‐mediated transformation (Yu et al., 2014).

Split‐luciferase complementation assay

The coding sequences of CATB and SRL10 (Table S4) were inserted into the N‐terminal (nLUC) and C‐terminal (cLUC) portions of firefly luciferase (LUC) respectively. Split‐luciferase complementation assay were performed according to the method described previously (Liang and Li, 2022) by using the resulting plasmids nLUC‐CATB and cLUC‐SRL10.

Co‐immunoprecipitation assays

The full‐length coding sequences of SRL10 and CATB (Table S4) were cloned into the pCAMBIA1300‐Ubi::GFP‐3 × FLAG vector and pCAMBIA1300‐Ubi::RFP‐HA vector respectively. pCAMBIA1300‐Ubi::CATB‐RFP‐HA was transiently co‐expressed with empty pCAMBIA1300‐Ubi::GFP‐3 × FLAG or pCAMBIA1300‐Ubi::SRL10‐GFP‐3 × FLAG in N. benthamiana. The detailed co‐immunoprecipitation assays were performed as previously described (Wang et al., 2020b).

Pull‐down assay

The coding sequence of SRL10 and CATB was cloned into pGEX‐4 T‐1 and pET28a respectively (primers used are listed in Table S4). The GST‐SRL10 fusion construct was transformed into Escherichia coli BL21 (DE3), and the recombinant protein GST‐SRL10 and His‐CATB were purified using a GST‐tag Protein Purification Kit (Beyotime) and a Ni‐NTA 6FF Sefinose (TM) Resin Kit (BBI) according to the manufacturer's protocol respectively. The detailed pull‐down assays were performed as previously described (Wang et al., 2020b).

Cell‐free degradation assay

Purified recombinant proteins GST‐SRL10, His‐CATB and total proteins from 20‐d‐old seedlings of WYJ7 were used for cell‐free degradation assay, and was carried out as previously described (Bello et al., 2019; Lv et al., 2014). The dilution for the mouse anti‐GST antibody (TransGen Biotech) and the mouse anti‐His antibody (TransGen Biotech) was 1:5000. Reactions were terminated at indicated time points, and the protein abundance was visualized via immune detection against anti‐His and anti‐GST. Total proteins from NIL‐SRL10 ^Hap1 and NIL‐SRL10 ^Hap3 were used for thermal stability analysis of protein and incubated under 28 °C and 42 °C for 0, 30, 60 and 90 min respectively. The dilution for the rabbit anti‐SRL10 antibody was 1:10000.

Bioinformatics analysis

RNA‐seq and small RNA‐seq analysis with three biological replicates was performed by Biomarker Technologies (http://www.biomarker.com.cn/) and Novogene (https://www.novogene.com/) respectively.

Single nucleotide polymorphisms (SNPs) in the genomic sequence of SRL10 were extracted from 3025 sequences found on the Rice SNP‐Seek Database (https://snp‐seek.irri.org/)(Mansueto et al., 2017). The three main SNP genotypes were extracted for haplotype phenotypic analysis, and the rice materials were divided into three main subgroups: aus, indica and japonica.

Construction of near‐isogenic line

Based on the BC₄F₆ chromosome segment substitution population of CJ16B which had experienced four generations of successive inbreeding with C84 as the recurrent parent, the lines carrying SRL10^CJ16B (CSSLs‐SRL10^CJ16B) between marker C10‐2 and C10‐3 were screened out. The near‐isogenic lines NIL‐SRL10 ^Hap1 (carried CJ16B) and NIL‐SRL10 ^Hap3 (carried C84) were constructed by using CSSLs‐SRL10^CJ16B backcrossed with C84 for two generations followed by inbreeding, then used for the treatment and yield test.

Quantification and statistical analysis

Quantification analyses on all the measurements were conducted in GraphPad Prism 8. Data are presented with mean ± standard deviations. Significant difference was examined by Student's t test and Duncan's multiple range test.

Author contributions

QQ and GZ designed the research. JW, JX, LW, MZ, JN, MC, XL, ZW, XL, JC, YL, ZZ, DZ, JH, LZ, GD, DR, ZG, LS, QZ, QL, LG and SY performed the research. QQ, GZ, JW and JX analysed the data. GZ, JW and JX wrote the paper.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

Figure S1 Transcriptomic analysis in leaves of WYJ7 and srl10.

Figure S2 Overexpression lines of SRL10.

Figure S3 Confirmation of the CRISPR/Cas9 transgenic lines of SRL10.

Figure S4 Phylogenic and conservative analysis of SRL10.

Figure S5 Scanning of interacting proteins of SRL10.

Figure S6 The relative expression level analysis of miRNAs in plants.

Figure S7 Subcellular localization of CATB and bimolecular fluorescence complementation assay.

Table S1 Genetic analysis of F₂ population of hybrid combination.

Table S2 Primers used for map‐based cloning.

Table S3 Primers used for sequencing.

Table S4 Primers used for vector construction.

Table S5 Primers used for qRT‐PCR.

Table S6 Accession lines of different haplotypes used for analysing of high temperature response and 2 years' heading date of these lines.